Automated Testing Gains Ground Using Robotics

Driven by advances in machine learning, using robotics for automated testing of software and hardware is becoming increasingly common. In fact, some IT experts believe that robots are on the verge of running sophisticated self-learning and self-healing tests – even  tests replicating the advantages that human judgment brings to the table. 

“Automated testing is key for companies to achieve their business goals of time-to-market with optimized cost and quality,” according to Tom Smith of DevOps Zone in a piece that surveyed IT professionals. “An automation-first culture will become pervasive, and the traditional approach of adding to testing debt will diminish over time.”

Advances in artificial intelligence and machine learning, Smith says, are expected to produce robots that have the cognitive abilities to self-learn, self-heal and self-adopt.

IT executives expect to see advances in artificial intelligence-powered cognitive vision toolsOpens a new window to emulate what human eyes process as visual differences in the testing process.

Automated testing is being used in both software and hardware tests. Lufthansa Technik of Hamburg, for instance, has automated cockpit inspections which the company says have become more efficient, thorough and rigorous.

“This fully automated procedure allows us to ease the burden on our colleagues in the workshops and reduce the testing effort by one to two hours per component,” saysOpens a new window Florian Sell, a senior engineer handling automated testing at Lufthansa Technik. “For example, we now have physical threshold values for the brightness of LEDs. And with the help of data mining, we can determine exactly when an LED has to be replaced.”

One of the biggest challenges for executives is determining how to best use people in robot testing. Some IT leaders say human judgment remains necessary in operations, but they also admit that mistakes are made sometime.

Others argue that even with advances in technology, human judgment will continue to be required.

“Sure, one day, machine learning, AI and robotics might get to a point where they’re completely replacing wholesale occupations or jobs,” saysOpens a new window  Patrick Kallerman, research director at the San Francisco Bay Area Council Economic Institute. “Nearer term, we’re going to see a substituting of tasks. The machines will do a percentage of tasks, and you’ll be freed up to do the rest.”

Relatively new companies are selling cloud-based testing platforms that use artificial intelligence. One such software company, Functionize Inc. of San Francisco, operates a machine-learning platform for clients to create their own tests, run them across all major browsers and then view the test results.

Functionalize uses advances in machine learning, such as Natural Language ProcessingOpens a new window , to make its platform accessible to clients regardless of their technical skills. Functionalize has partnered with Google to take advantage of cloud servicesOpens a new window , so that clients can test from any city and in a range of bandwidth restrictions. Both services allow the testing to simulate the experience that users from other countries might have on their sites.

Functionalize sells the ability of its testing platforms to automatically self-maintain as its key feature.

“One of the biggest timesinks with automated testing is test maintenance – the slightest change to a style sheet or a simple change to the page layout breaks the script.,” writesOpens a new window Paul C on Functionalize’s website. “Our tests are completely self-healing. Each action is fingerprinted by the system, which uses machine learning to establish what the action is actually doing.”

Another startup, Mabl.com of Boston, has developed a function that identifies future potential issues, using machine learning and artificial intelligence advancesOpens a new window to replicate the advantages that human judgment brings to the testing process.

Another advantage of Mabl is that it collects a vast amount of data from a range of sources, using the increased processing and analytical power of the cloud.

“As you run tests in Mabl, it collects rich data that spans not just test output logs but also HTML data, images, screenshots of the state of pages, and errors coming out of Chrome and other browsers,” writesOpens a new window Joe Colantonio on his website on automation technology advances. “All this test data is fed into a machine-learning pipeline. This allows the tool to look at all the different data sets of what the application is doing, and what is normal about the application for each step in each test.”